Bitcoin burns electricity to secure transactions. Rhea burns tokens to verify knowledge.
Both produce immutable artifacts. Both share them with the world.
💎 wait like bitcoin but for science?
Bitcoin miners burn electricity to solve math puzzles → get BTC (proof they did the work).
Rhea burns AI tokens to verify claims through independent models → gets truth gems (proof the claim survived scrutiny).
Both are immutable, verifiable, public. Nobody can fake a gem — the hash chain proves the work happened.
💎
Aletheia — Truth Gems
Every verified claim becomes a gem: SHA-256 hashed, hash-chained, graded A/B/C by consensus strength.
5 gems mined so far. 36K tokens burned to produce them.
Each gem records which models agreed, where they diverged, and what evidence survived.
What it is: The entangled limit of everything computationally possible —
following all possible rules in all possible ways. Unique, inevitable, formally necessary.
No choices needed to define it; it just is, like 1+1=2.
Why it matters: We are inside it. We never see the Ruliad from outside —
we sample tiny slices using our bounded perception. For “observers like us”
(computationally bounded, coherent, persistent), the laws of general relativity and quantum mechanics
inevitably emerge. Physics and mathematics are the same object viewed from different angles.
Rulial space: Like physical space, but for description languages.
Moving in it = changing how you think. Concepts are “rulial particles” —
stable structures that survive transport between different minds.
Different consciousnesses are at different positions in rulial space;
that’s why you can’t fully “get inside” another mind.
The punchline: Computational irreducibility means no shortcuts —
you must walk through time step by step. The universe runs every possible rule;
we just perceive the slice consistent with what we are.
∞-groupoid structurecausal invarianceWolfram 2021
BTC: electricity → SHA-256 → immutable block → shared ledger Rhea: tokens → multi-model consensus → truth gem → proof chain Both burn real resources to produce tamper-proof public artifacts.
The mining tools
Everything that turns questions into verified gems.
🧐 ok so what are these tools
These are the pickaxes. The gems above are the gold.
Each tool here is a different way to dig — ask 5 scientists, send a red-teamer to attack your idea,
or set a robot to keep digging until it hits 90% consensus. You pick the tool, Rhea does the work.
⚖
Tribunal Consensus
Query multiple AI models independently. They don’t see each other’s answers.
Agreement = signal. Divergence = where the lie hides. 3-model, 5-model, or ICE deep verification.
🤖
Sceptic Mode
Adversarial verification. One model attacks the claim, others defend. Stress-test before you trust.
🎯
Scheduler Looper
Set your quality bar (90%+), submit a claim, walk away. Auto-refining rounds until models converge.
⚡
Workflow Engine
Chain tribunal calls, proof storage, and notifications into automated pipelines.
🔑
BYOK
Your API keys. Your providers. $0 platform fee. You own the infrastructure.
Anthropic
OpenAIGoogle
Meta
Groq
Cerebras
Redis
Oracle
Firebase
Wolfram
NDI
Full pipeline in <5 MB. Free.
Type a claim anywhere. The keyboard sends it to Rhea’s servers. You get back verified science.
⌨️ wait the keyboard does what
Your phone keyboard becomes a portable lie detector.
Type “vaccines cause autism” and 5 AI brains check it before you can say “peer review.”
All in an app smaller than a selfie. Free.
Drug Discovery
> “Aspirin inhibits COX-2 selectively at low doses” ▶Agreement: 34% — COX-1 and COX-2 both inhibited non-selectively ✓ Sceptic found 3 contradicting sources
Caught a common pharmacology misconception in 4 seconds
Molecular Biology
> “CRISPR-Cas9 has no off-target effects in vivo” ▶Agreement: 12% — off-target cleavage well documented ✓ 5 models unanimous: claim is dangerous misinformation
Prevented a false safety claim from entering a grant proposal
Grant Writing
> “Lipinski’s Rule of Five predicts oral bioavailability” ▶Agreement: 78% — valid heuristic, not absolute ✓ Nuance preserved: “predicts drug-likeness, not bioavailability directly”
Refined a claim from “mostly right” to “precisely right”
⌨️
Rhea Keyboard
iOS Extension • <5 MB • zero dependencies • FREE
A thin client that sends claims to the full
ComfyUI verification pipeline on Rhea’s servers. Type in any text field, tap verify.
The server runs the complete node graph. Result drops back in seconds.
100 free queries/monthNo GPU neededWorks in any app
Cross-platform web UI — Windows, Linux, any browser
One clipboard. Every device.
Copy on your Mac. Paste on Windows. Replay anything. Secrets auto-expire.
📋 passing notes in class but cooler
Remember passing notes in class? This is that, but between your phone, laptop, and grandma’s PC.
And the note self-destructs if it has a password on it.Spy stuff, basically.
📋
Cross-Device Sync
Copy a URL on your phone, paste it on your laptop.
Real-time SSE push — no polling, no delays. Works across macOS, iOS, Windows, and Linux.
⏰
History & Replay
Every clip saved with timestamp, source device, and app.
Search your clipboard history. Pin important clips. Replay any entry from the last 30 days.
🔐
Privacy by Default
Auto-detects passwords, API keys, and credit card numbers.
Sensitive content gets a 30-second TTL and E2E encryption. You control what persists.
macOS • iOS
Built into Rhea app
⇄
☁️
Rhea Cloud
E2E encrypted
⇄
🪟
Windows • Linux
pip install rhea-clipboard
Set the bar. Walk away.
Consensus loops run until your quality threshold is met. Like ComfyUI, but for truth.
🔄 a washing machine for facts
Like a washing machine for facts. Throw in a dirty claim, set the “how clean” dial to 90%,
press start. It keeps washing until it’s sparkling.
You don’t even have to watch. Go eat a sandwich.
🎯
Target Consensus
Set your quality bar — 85%, 90%, 95%.
The scheduler runs tribunal rounds automatically until models agree at your threshold.
🔄
Auto-Refine
Each iteration learns from divergence.
The prompt sharpens itself, targeting exactly where models disagree. Convergence accelerates.
🛠
Three Modes
Local: fast parallel consensus.
Sceptic: adversarial attack each round. ICE: iterative deep critique with chairman synthesis.
Submit
Your claim enters the pipeline
Tribunal
k models deliberate independently
Converge
Meet threshold or refine & iterate
Proof
Stored in Aletheia with provenance
POST/workflows/scheduler/looptarget_agreement:0.9
Start free. Scale when ready.
100 credits on signup. Bring your own keys for zero platform cost.
💰 the money part
100 free fact-checks when you sign up. After that — bring your own AI keys and pay $0 forever.
We literally built a business model that lets you not pay us. Your move.
Free
$0
Everything you need to verify claims with multi-model consensus.
95% becomes your credits instantly. 5% funds the impact above. No middlemen. First payment auto-creates your profile —
patron key derived from tx hash, redeemable anytime.
Available everywhere
One account. Every platform. Native experience.
📱 runs on everything except toasters
Works on your iPhone, your Mac, your browser, your terminal.
Basically if it has a screen and isn’t a toaster, Rhea runs on it.One account, everywhere.
💻 one command, that is it
Copy this, paste it in your terminal, boom — you just asked 3 AIs if your science homework is right.
Your teacher can’t do that.No SDK, no library, no drama.
The infrastructure owner controls who's admin, not the application.
You own your data, your models, your keys. Rhea serves you — not the other way around.
We use cookies for authentication and analytics. See our Privacy Policy.